Architecture Overview

Architecture Overview

As mentioned previously, the thin client architecture is very similar to the architecture of Internet applications. It is based on a three-tier architecture, as depicted in Figure 11.1.

Click To expand Figure 11.1: Wireless Internet architecture.

The following are the key components of the thin architecture:

  • Thin client. The client is a microbrowser focused on the presentation of the markup language. It is the part of the application that the end user sees, although it does not execute any of the application logic. The design of the client user interface is crucial to the overall success of the application.

  • Middle tier. The middle tier is where most of the work is performed. A couple of main components run in this tier: the Web server and the wireless application server. The Web server receives the inbound HTTP requests and routes them to the wireless application server. The application server then takes these requests and executes the appropriate logic. This execution may include session management, content management, as well as integration to the back-end system.

  • Back-end system. The back-end system is where the data and related services reside. This system may be a relational database, an email server, business applications, an XML content source, or any number of other enterprise systems. In any case, it is important that this data can be reached, so that it is available to the mobile worker.


The term wireless application server refers to the middle-tier server that contains both an application server and a framework designed for wireless content delivery. These two components may be supplied either together or independently.

Figure 11.1 shows a very simple three-tier architecture. In reality, several other layers need to be addressed: the security layers, including firewalls, as well as the actual location of the wireless networks and related wireless gateways. In the rest of this section, we will cover each part of the architecture in more depth.


The client interface for thin client applications is an Internet browser. In the desktop world, browsers have many advanced capabilities, with sophisticated user interfaces, often supporting client-side scripting languages such as JavaScript or VBScript. Most desktop browsers also have advanced graphics and multimedia support. Many of these features, however, require high bandwidth and powerful client-side processors, a luxury not available on most wireless devices. The client for most wireless applications is not nearly as sophisticated. The browsers have limited screen size, restricting the richness of the user interface. In addition, there is no de facto standard for wireless browsers, meaning that, to reach a broad audience, application providers have to support a variety of browsers and many markup languages.

With these challenges in mind, we will take a look at the various aspects related to the client layer of wireless Internet applications.

User Interface

The user interface may be the most important part of a wireless Internet application. People may disagree, but if users cannot quickly find what they are looking for, they will stop using the application, in which case, everything else becomes irrelevant. The problem is, even the best-designed user interfaces can cause user frustration. Trying to cram useful data and navigation controls onto a small screen such as a cell phone is a daunting task. Even when the screen size is larger, such as on a PDA, the task is not much easier. Much care has to be taken to ensure that the required information is easy to find, without having to navigate through a dozen screens to get it.

One of the difficulties in this endeavor is that the way we navigate desktop Web applications is fundamentally different from how we navigate wireless devices. A wireless device is not commonly used for general information gathering and Web surfing; rather, the wireless device is used for a specific purpose, and the user interface should make accomplishing that purpose as efficient as possible. Therefore, applications have to be targeted for specific tasks, such as placing an order or looking up a defined set of information. Examples in the corporate space include a salesperson looking up price lists or inventory levels, or a field service employee finding his or her next work order. In the consumer space, such a task may involve looking up a stock quote, purchasing a movie ticket, or viewing the latest sport scores. In all these examples, the application is being used for a specific reason; the faster users can accomplish their goal, the more satisfied they will be. The most effective user interfaces do one job and do it well. (In Chapter 12, "Thin Client Development," we will take a closer look at some design patterns that are useful for creating effective wireless Internet user interfaces.)

Browsers and Content Types

Another challenge in creating wireless thin client applications is the variety of markup languages and microbrowsers currently being used. Whereas in the wired world you can address an overwhelming majority of your target audience by creating HTML applications that run in Microsoft Internet Explorer and Netscape Navigator, the same is not so with wireless Internet applications.

Each device typically has its own browser or, in many cases, set of browsers. Each browser understands a particular markup language. The common markup languages include Wireless Markup Language (WML) as defined in WAP 1.x; XHTML, as defined in WAP 2.x; Handheld Device Markup Language (HDML), which is a predecessor to WML; and the subsets of HTML, such as Compact HTML (cHTML). Each of these markup languages has nuances that have to be addressed. Figure 11.2 shows the common microbrowsers on a variety of device classes.

Click To expand
Figure 11.2: Common microbrowsers— (a) Openwave browser, (b) Go.Web browser on RIM 957, (c) Pocket Internet Explorer, (d) Palm Web Clipping.

Moreover, each browser handles markup languages differently, meaning you have to test each browser/markup language combination to be sure that the application is displayed correctly. For example, WML content may display correctly on a cell phone running the Openwave browser, but not function properly on a RIM device with a GoAmerica Go.Web WML browser. For such a situation, you have to create content with slight variations to accommodate different client browsers. The good news is that many wireless application servers incorporate this functionality, so it does not have to be handled manually. (In Chapter 13, "Wireless Languages and Content Generation Technologies," we take an in-depth look at the leading wireless markup languages and content-generation techniques for creating device-specific applications.)

Finally, there are voice browsers that use VoiceXML as the markup language. These applications do not display content visually, but rather speak the response over any telephone; the interaction with the system is through voice as well. (Note that because voice-based applications do not use the typical wireless gateways, they are covered separately, in Chapter 15, "Voice Applications with VoiceXML.")

Wireless Networks

In order for thin client applications to receive data from the server, a wireless network connection is required. This is different from smart client applications, which store data locally, allowing for offline access. With thin client applications, each request goes over a wireless network to a wireless gateway, at which point it is transferred to an HTTP request and forwarded to the Web server/wireless application server.

In most cases, the browser takes care of the communication with the wireless network. This saves developers from having to know the underlying network protocol, allowing them to focus on the application logic. This is a significant advantage, as a large number of wireless network protocols are being used. A complete step-by-step analysis of a wireless request procedure is included in the section Processing a Wireless Request later in this chapter.


In wireless thin client applications, most of the work is performed using middleware specialized for wireless applications. In this section we take a look at the two main types of servers: wireless application servers and wireless gateways.

Depending on the corporate requirements and resources, these servers can either be hosted by a third-party vendor or be kept on premise. In the most common configuration, the wireless service provider hosts the wireless gateway, and the wireless application server is located within the enterprise.

Wireless Application Servers

Wireless application servers provide the core infrastructure upon which wireless Internet applications are built. They are the engine driving the wireless Internet architecture, providing many of the core feature requirements for this type of application. To meet this rather tall order, wireless applications servers must have a strong server foundation on which the wireless components can be built. For this reason, many vendors have chosen to use J2EE application servers as the underlying technology for their wireless offerings. Using J2EE enables wireless vendors to focus on the core requirements for wireless applications, while leaving the generic server capabilities to the application server vendors.

Initially, many vendors created their own proprietary servers, which were often targeted at a limited number of platforms, thus requiring the developer to use proprietary markup languages and APIs. Realizing that developers preferred to use standards-based technology, most vendors since have ported their frameworks and tools to standard J2EE servers produced by BEA Systems, IBM, Sybase, and Sun Microsystems. In doing so, wireless application server vendors have increased the number of platforms they can support and have reduced the complexity of their offerings. Later in this section, we will investigate the core features of applications servers and how they relate to the J2EE standard and the Microsoft .NET platform. But first we will examine the features provided by commercial wireless server products that are designed for the unique requirements of wireless applications.

The first implementations of many wireless server products were targeted at specific content/device/network combinations. An example would be WML content for the Openwave browser on Nokia devices over a GSM network. While this generated a positive result for the specific combination being targeted, it did not scale well for the multitude of browsers, devices, and networks that were being used as wireless technology became more mainstream. In order to succeed in this rapidly changing marketplace, wireless server technologies had to adapt. This involved creating frameworks that allowed for the development of a variety of wireless applications that could be deployed to a wide range of devices over an assortment of wireless networks.

These frameworks became the basis for the wireless server products on the market today. Most commonly built using J2EE technology, including Java servlets, JSPs, and EJBs, these frameworks provide the core features that are required for wireless application development. These features include device and browser identification, content transformation, voice technology integration, session management, and enterprise data integration. In addition, framework-specific development tools are often available to lower the learning curve associated with using this new technology. Each feature serves a very specific purpose that contributes to the overall success of the wireless application being developed.

A few features on this list generate the most interest. The device and browser identification capabilities are key to providing the appropriate content to the user. By identifying the client device and browser, we can determine the required markup language, the size of the user interface, the types of graphics supported, and the preferred language, among other things. Once we know these factors about the client, the content can be created accordingly, using a variety of techniques and technologies. (In Chapter 13, we take a closer look at each of the leading markup languages, as well as the most common content-generation techniques. In Chapter 14, "Wireless Internet Technology and Vendors," we take a closer look at the features of wireless applications servers and provide summaries of the commercial solutions available.)

Application Server Basics

Here we will take a quick look at the basic services that application servers provide as they apply to wireless Internet applications. (An in-depth look at the capabilities of an application server is beyond the scope of this book.)

The application server is the cornerstone of the distributed Internet architecture. It connects different components of complex applications, while allowing the application to remain modular. It provides the glue to pull together enterprise data, business logic, and the client presentation layer. All of these capabilities are provided in such a way that frees developers from being concerned with the implementation details, allowing them to focus on building the application logic and user interfaces. In general, application servers provide the following three distinct layers:

  • Web integration

  • Component transaction server

  • Enterprise connectivity

Because many commercial wireless servers utilize other application servers as their base platform, we also have to examine the core technology on which they are built. That endeavor involves investigating the core features of the two leading specifications implemented by application server vendors: J2EE and .NET. They both provide the same core functionality, but in significantly different ways; hence, we will review each technology individually.


The Java 2 Platform, Enterprise Edition (J2EE) is a specification delivered by Sun Microsystems using the Java Community Process (JCP), which is used to develop and revise Java technology specifications with input from the international community of Java technology experts. In other words, it allows the greater Java community, rather than a single entity, to govern the J2EE specification (and other Java specifications).

The J2EE specification defines a standard platform for building multitier enterprise applications. It builds upon capabilities provided in the Java 2 Platform, Standard Edition (J2SE), such as the core Java libraries, the promise of code portability, and Java Database Connectivity (JDBC). In addition, it defines an enterprise component model in Enterprise JavaBeans (EJBs), Internet capabilities through Java Servlets and JavaServer Pages (JSPs), and XML integration support. Vendors cooperate on developing the specification, then compete on the implementation. This allows a broad range of enterprise systems to use a single standard, allowing for portability between technology vendors and preventing corporations from being tied to a single vendor. Many corporations have desired this type of specification for some time.


Keep in mind is that J2EE is not a product; it is a standard to which products can be written.

Wireless Internet applications usually take advantage of four main J2EE technologies: EJB, Java servlets, JSPs, and JDBC. They may also use others such as Java Naming and Directory Interface (JNDI), JavaMail, and Java Transaction API (JTA), but usually to a lesser extent, so we will not discuss them here.


The Java Message Service (JMS) is also an important technology for mobile and wireless applications. It was discussed in more depth in Chapter 5, "Mobile and Wireless Messaging," and is mentioned again later in this chapter in the Messaging Servers section.

Let's take a look at each of these four technologies in relation to the functionality they provide to Internet applications. Figure 11.3 provides a logical diagram of how these core layers may be implemented. As you can see, servlets and JSPs act as the main interface between the client applications and the enterprise business logic contained in Enterprise JavaBeans. The EJBs can then communicate with the enterprise data source using JDBC or other adapters.

Click To expand
Figure 11.3: J2EE architecture.

Java servlets and JavaServer Pages provide the capability to create dynamic, data-driven content. Java servlets enable developers to easily implement Java server-side code that takes advantage of the other J2EE APIs. They act as an interface between the client application and the enterprise business logic. JSPs provide the same interface, but using a scripting language such as HTML or WML. The JSP specification supports static templates, dynamic markup language generation, and custom tags. Both Java servlets and JavaServer Pages play a prominent role in most wireless application server products.

Business logic is encapsulated in Enterprise JavaBeans components. EJBs make it possible for application developers to create reusable segments of logic, without having to be concerned with the underlying application complexity. The EJB container within the application server handles the various details such as threading, component pooling, and garbage collection. Developers can implement two types of EJB components: Session Beans and Entity Beans. Session Beans typically contain the business logic that maintains workflow and transactions. Entity Beans represent collections of data. They encapsulate data sets and enable operations that can be used to interact with that data. Both component types allow for rapid development and deployment of enterprise logic. Most wireless application servers support the use of EJBs for business logic, but they often do not utilize them for the core wireless capabilities.

The final layer is the enterprise integration layer. If the data resides in a relational database, then JDBC will most likely be the technology used to interact with the data. Most enterprise data vendors have JDBC drivers for their offerings, providing a standard way to manipulate their databases. JDBC offers support for user-defined types, rowset manipulation, connection pooling, and distributed transactions.

If the data does not reside in an enterprise database, then a custom adapter most likely will be required. Depending on the data source, a commercial adapter using EJB technology may already be available. In most cases, these adapters are built to the EJB specification, so they should be able to work in the application server you are using (assuming it is J2EE-compliant). If a commercial adapter is not available, then you may be required to implement an adapter in-house. This undertaking may not be as difficult as it seems. Many enterprise applications have defined APIs that you can use to access the systems, often available for the Java programming language. Additionally, the Java Connector Architecture (JCA) provides an industry standard way to integrate with legacy or nonstandard data sources from J2EE applications.


All of the leading J2EE application server vendors are compliant with the J2EE specification. This means that J2EE components such as Java servlets, JSPs, and EJBs written to the appropriate specification should be portable across application server offerings. We say "should," because in reality, testing is usually required on each platform to adjust for vendor-specific nuances.

Microsoft .NET

Microsoft .NET is the overarching technology around which Microsoft is building its entire platform. This includes its Visual Studio development tools, programming languages, mobile operating system, messaging architecture, and database servers. In the future, most Microsoft products will be part of the .NET initiative. A major component of .NET is XML Web services. Web services are geared to allow applications to communicate and share data over the Internet, independent of the operating systems or programming language being used.

To help highlight the .NET platform, we will compare the J2EE standard and the .NET products. Notice that J2EE is a standard to which products can be developed, whereas .NET is a set of products offered primarily by Microsoft. We will take a look at these products in relation to how they compare with the J2EE standards.

Let's start with the programming language of choice, C#, the base language of the .NET platform (although other languages can be used as well). In fact, Microsoft is promoting the Microsoft Intermediate Language (MSIL), a set of byte codes into which any supported language (such as VB.NET and C#) can be compiled. These byte codes are then interpreted and translated into native executables using the common language runtime. A strong correlation exists between MSIL and the common language runtime and Java byte codes and the Java Runtime Environment (JRE).

Beyond the programming language, the three main aspects of .NET that we will investigate are the Web integration layer, the component transaction layer, and the enterprise integration layer. Figure 11.4 illustrates the main technologies that are of interest to us in the .NET platform. These include (Active Server Pages) ASP.NET, .NET-managed components, and ADO.NET.

Click To expand
Figure 11.4: .NET platform architecture.

As the name suggests, Active Server Pages .NET are a scripting technology similar to JavaServer Pages. They provide dynamic content to a variety of clients, including desktop clients using Windows forms and browser clients using a variety of markup languages including HTML and WML. As depicted in Figure 11.4, ASPs provide the integration layer between HTTP-based clients and the server-side business logic.

Business logic is encapsulated in .NET-managed components. These are the successors to COM and COM+ components and provide interoperability with them. Essentially, .NET-managed components provide the business logic and data processing layer of the .NET platform, similar to what EJBs provide to J2EE platforms. The business logic for .NET-managed components can be written in a variety of programming languages, since the language will be compiled into the MSIL and executed using the Microsoft common language runtime.

The final layer that we will look at in .NET is the enterprise integration layer. ADO .NET is the technology that provides connectivity to enterprise data sources. It is the latest in a long line of database access technologies that started with Open Database Connectivity (ODBC) many years ago. Since then we have seen OLE DB and ActiveX Data Objects (ADO). One of the main differences of ADO.NET over its predecessors is that it is not as database-centric. ADO.NET is required to support other data sources, as it is the main integration layer between the .NET-managed components and enterprise data. For instances where ADO.NET does not provide adequate enterprise integration, custom adapters can be created using the .NET-managed component architecture. In addition, because .NET is focused on Web services, a variety of XML-based protocols are available for data integration.

J2EE versus .NET: Summary

Debate is ongoing as to which platform is best: J2EE or .NET. Resolving this debate is not easy, as arriving at an answer involves detailed technical and business analysis. The capabilities of both platforms are comparable, but the ways they are implemented are distinctly different. J2EE is a standard that is widely supported throughout the software industry. There are well over 50 vendors that offer J2EE support on a variety of software and hardware platforms. This results in a wide range of companies providing tools, servers, and other products related to J2EE technology, giving the application developer many options for developing and deploying J2EE applications.

.NET, on the other hand, is a product suite offered by Microsoft on Windows platforms. It provides strong tool and server offerings, but not much flexibility outside of Microsoft's product realm. Companies will either see this as a limitation, since they are locked into Microsoft technology, or as a benefit, since it does provide a tightly integrated solution. Having everything provided by a single vendor does have positive aspects: You will have some assurance that the various technologies will work well with each other and with the operating system used for deployment.

Choosing which platform to use comes down to your specific application and business requirements. Both J2EE platforms and Microsoft .NET are enterprise-class offerings. If you are looking for an open solution, with wide industry and vendor support, J2EE is most likely a better choice for you. If you are primarily using Microsoft technology, and are looking for a well-integrated solution from a single vendor, the .NET product suite should suit your needs well.

Wireless Gateways

Wireless gateways are the link between wireless and wired networks, providing integration between wireless communication protocols and the enterprise. This allows all wireless applications to access enterprise applications and data through the same entry point, independent of the wireless technology being used. Wireless gateways allow users with varying needs (devices, applications, coverage) to select the best network for their situation.

As discussed in Chapter 3, "Wireless Networks," wireless networks use many different protocols for communication, often not based on IP. This means that wireless gateways have to obtain a direct connection to the particular wireless carrier for communication, since it cannot travel over the Internet, which is IP based. With this connection in place, the gateway then transfers the information from the proprietary wireless protocol to an IP-based protocol, such as TCP/IP or HTTP(S) allowing for direct communication to the enterprise system. In this way, the wireless gateway is able to provide connectivity from a wide range of wireless networks to enterprise systems.

Due to the complexity and time required to implement and maintain the connections, wireless gateways are usually hosted by third-party vendors outside corporate firewalls. Here's a concrete example. Let's assume the client is a Go.America browser running on a RIM device in North America. The wireless protocol being used is Mobitex, which is not IP-based. The wireless application server is listening using HTTP for any incoming requests. The browser on the RIM device is not using HTTP, so it cannot send a request directly to the enterprise wireless server. At some point between the client and the wireless server, the request has to be adapted from the Mobitex protocol to HTTP. This is one of the roles of the wireless gateway. Next assume that a request is made by the user to retrieve content from Once made, the request is transferred wirelessly to a base station in the user's vicinity. From there, it will be routed using a direct wired connection to the network operation center (NOC), where the wireless gateway resides. The wireless gateway then takes the request, coverts it to HTTP, and forwards it to the appropriate enterprise Web server. The Web server can then output the appropriate WML content and send it back to the wireless gateway, where it is changed back to the Mobitex protocol and sent back to the user. In this scenario, the wireless server and the gateway are located on separate servers in different locations. (Note: Content transformation capabilities are not always handled this way. In many cases, they are executed by the wireless gateway as well.)

Other features offered by the wireless gateway include optimized communication streams, support for IP over non-IP networks, WAP push messaging, and enhanced security. For some of these features, an application must be deployed to the client device, but not always. It depends on the specific feature, as well as the vendor supplying the technology.

One common wireless gateway is a WAP gateway. WAP gateways provide the link between clients using WAP and enterprise systems. Later in this chapter, in the section called Wireless Application Protocol (WAP) Overview we will take a closer look at the role of the WAP gateway within the WAP architecture.

Messaging Servers

Recall that in Chapter 7, "Smart Client Overview," we looked at messaging as it pertained to smart client applications. Here we focus on the aspects that are relevant to thin client applications. Included in this category are the Short Message Service (SMS), HDML notifications and alerts, and WAP Push. (Note: Wireless email will not be covered here, as it is discussed within the context of mobile information management in Chapter 16, "Mobile Information Management.")

Let's start by examining the types of messages that thin client applications can receive. As you may recall, the only application on the mobile device is the micro-browser. All of the input and output for wireless Internet applications goes through the browser interface. This is often the case for thin client messaging. There is no persistent data on the device, meaning the messages sent are intended to be viewed by the user, not to be consumed by an application, as in smart client messaging. The most common form of text messaging is SMS, particularly in Europe and Asia, and more often now in North America.

SMS is based on a packet-based architecture that sends data over the control channel. The messages are typically limited to 160 characters. Typical uses include weather reports, stock quotes, calendar reminders, and communication between two users. SMS messaging is controlled using SMS centers, or SMSCs. Each carrier typically has its own SMSC, with its own proprietary interface. In order for server applications to send messages using SMS, they must have interfaces to each of the SMSCs being used. Implementing this in-house can be a huge burden. For this reason, companies have created a common interface to the various SMSCs to make server integration easier. Using this type of service prevents you from having to get your own connection to each carrier, and from having to interface directly with each carrier's APIs.

The size limitation of SMS messages restricts the content being sent. Text is the only practical messaging format that can be sent in 160 bytes. Because of this limitation, sending more feature-rich messages using this protocol is not possible. The Enhanced Message Service (EMS) changes this. It adds powerful new functionality on top of the SMS standard. With EMS, mobile phone users can add some "life" to SMS text messages, in the form of pictures, animation, and sound. This gives users new ways to exchange pictures and ringtones with other users. One of the great things about EMS is that it is based on SMS standards, meaning that wireless carriers can support EMS with little effort.

In addition to SMS and EMS, WAP 2.x provides a new messaging service called Multimedia Message Service (MMS) that allows for even richer messaging. MMS is aimed at providing a feature-rich messaging solution, in both instant delivery mode (similar to SMS) and with store-and-forward queuing (similar to email). In the future, MMS is expected to become the preferred messaging protocol for mobile users, since there are virtually no limits on what can be sent using it. An MMS message can contain graphics, data, animation, images, audio clips, voice transmissions, and video sequences. That said, it's important to point out that MMS messages are much larger than SMS messages, so MMS will not find widespread use until wireless bandwidth increases.

Outside of the SMS world, browser alerts and notifications are also used for messaging. They provide an alternative means of sending information to wireless Internet users via the microbrowser. In WAP 1.2, the concept of WAP Push was introduced. WAP Push enables a content server to push content to a WAP 1.2-enabled handset by sending a specially formatted XML document to the Push Proxy Gateway, which in turn forwards it to the handset.

Another form of quick-delivery text messaging is instant messaging (IM). Instant messaging from companies such as AOL, Microsoft, and Yahoo! is common on desktop clients, and is quickly moving into the wireless world. One difference between instant messaging and SMS is that IM uses the Internet to communicate. One of the challenges of using wireless IM is that no standard is in place, so IM vendors have their own proprietary interfaces in their IM servers. Though this situation sounds similar to that of SMSCs, with all the challenges that entails, actually it is quite different. With SMSCs you have to integrate with proprietary interfaces, over proprietary networks, whereas with IM, you are communicating over the Internet, meaning you do not have to have a fixed connection to the IM servers. The only level of enterprise integration that is required is communication with the various servers.


For the purpose of this chapter, we focus on the text-messaging capabilities in thin client applications. Although more advanced messaging with multimedia is possible, it is not commonly used at this time due to the current state of wireless networks and mobile devices. For a more complete look at messaging as it relates to all forms of mobile and wireless computing, see Chapter 5, "Mobile and Wireless Messaging."